Introduction

In recent decades rat liver cell cultures have been utilized by many investigators for various purposes in liver cancer research. Although it was generally believed that in vivo most of the hepatomas were derived from hepatocytes, none of the hepatocyte cultures used for in vitro studies could survive longer than several weeks keeping their normal functional activities. On the other hand, rat liver epithelial cells (RLEC) could develop into long term cell lines or strains. While none of them showed all the hepatic functions, they maintained some of them, including hormone responses to synthesize tyrosine aminotransferase (TAT).

In vitro studies with RLEC have afforded useful information to hepatocarcinogenesis research. In some experiments, spontaneously transformed RLEC were capable of producing carcinoma like tumors 1, while chemically transformed RLEC resulted only in a small portion of hepatocellular carcinomas associated with a wide range of other tumor types 2. The high heterogeneity of tumors formed by transformed RLEC may to some extent reflect the original relationship between hepatocytes and culturable RLEC. Therefore studies with RLEC may also be helpful in finding out the real progenitor(s) of hepatocellular carcinoma in vivo.

It has become apparent that oncogenes might be involved profoundly into liver cancer formation. Among these oncogenes, activated ras and/or myc genes were detected in chemically induced rat liver cancer 3 and particularly in human primary hepatoma 4. With fibroblast system there is some evidence suggesting a cooperation of ras and myc genes during the malignant transformation 5. So far, to this point, studies with rat liver cells gave some controversial results. Simultaneous activation of ras and myc has been reported for several systems. But for myc gene, while some reports indicated that c myc expression was essential for transformation either being associated with early 6 or late stage 7, other data argued that elevated myc expression might be related to immortalization of RLEC rather than ras induced transformation 8.

Several laboratories have reported transformation of RLEC by EJ Ha ras gene 9, 10, 11 but the possibility of formation of hepatocellular carcinoma by these transformed cells through transplantation was not mentioned. And the molecular basis of this ras induced transformation remained to be elucidated. In addition, it would be interesting to investigate whether the integration of the exogenous ras gene into a specific host chromosome is a common phenomenon in the RLEC system,as has been reported previously 12.

In this paper we described the establishment of the rat liver epithelial cell line LW13. Subsequently, we transformed this cell line with the EJ Ha ras gene,aiming at analyzing the behavior of Ha ras and other oncogenes during the transformation process, mimicking in vivo hepatocarcinogenesis to some extent.

Methods and Materials

A. Establishment of LW13 cell line

The method used for the establishment of the cell line was basically similar to that described by G. M. Williams 23. Briefly, 10 d old male Wistar rats were cervically dislocated, the extirpated livers were washed, minced and digested by D Hanks' buffered 0.3% trypsin in flask with magnetic bar stirring at 37 C for 10 min. After decantation the trypsinization process was repeated twice. Combined cell supernatants were centrifuged, resuspended in HamF12 10%BS medium,and dispersed into culture flasks. After 20 min incubation at 37 C, the medium containing unattached cells was transferred to a second set of flasks for another 20 min incubation. Thereafter, still unattached cells in the supernatants were transferred to a third set of flasks and incubated again as described. The final supernatants were discarded and the flasks containing attached cells were covered with HamF12 10%BS medium, supplemented with insulin, glutamine, glucose and antibiotics. Subcultures were maintained in DMEM HamF12 10%BS (or FCS) mediurn. Cells were passaged when they were grown just to confluency. The medium was changed twice per week.

B. Characterization of the LW13 cell line

Slides for chromosome studies were prepared by standard techniques using hypotonic treatment of colcimid arrested cultures 13 and were analyzed after Giemsa trypsin banding. Determination of TAT activity was performed after the method of Granner 14. Examination of ultrastructures was carried out with glutaraldehyde and OsO4 fixed cells as described 15. Sections were examined with JEOL JEM100B transmission electron microscope (TEM), Measurement of growth curve was done by daily counting duplicate flasks passaged with an initial density of 2×105 cells/ flask.

C. Plasmid DNA

For transformation of the LW13 cells, a plasmid consisting of pSV2 neo sequences 16 and a 6. 6 kb EJ Ha ras transforming oncogene 17 incorporated into the BamHI site of pSV2 neo was employed:a 460 bp fragment of v Ha ras in plasmid BS9 33 was used as probe for the detection of transfected EJ Ha ras sequences; a 2. 2 kb SstI BamHI fragment generated from 6.6 kb EJ Ha ras, which could not be hybridized with endogenous rat cellular ras sequences, was used as a probe for in situ hybridization of chromosomes; the probe used for the detection of the expression of endogenous c myc gene was a 420bp pstI pstI fragment of mouse c myc exon 2.

Small and large scale preparations of plasmid DNA were carried out with alkali SDS methods similar to those described by Maniatis 18, 19. A Bio Gel A 150m column was used in some cases to separate plasmid from RNA. Raw DNA was further purified by repeated CsCl (Merck) centrifugation after Maniatis [2()] to obtain circular plasmid DNA for transfection.

D. Transfection of LW13 cells

The transfection of EJ Ha ras gene into LW13 cells was carried out using standard calcium phosphate precipitation technique. Briefly, 5 μg pSV2 neo Ha ras or pSV2 neo plasmid DNA and 10–20 μg high molecular weight salmon testis DNA were mixed, and incubated with cells (passage 25–30) grown to 50–80% confluency in 60 mm culture dish after precipitation with CaCl2 and HEPES buffered saline. Transfectants were selected by addition of the antibiotic G418.

E. Preparation of radiolabeled probes

Probes used for nucleic acid hybridization were labeled by incorporation of [α–32P] dCTP(NEN) with a nick translation kit (Amersham N5500), probes used for in situ hybridization were labeled by 3H dATP or 3H dCTP (Amersham) with the same kit or with the Amersham Multiprimer labelling Kit RPN1600Y. Radiolabeled DNA was isolated by Sephadex G50 chromatography.

F. In situ hybridization

The method was based on that of Chandler 21. Normally, 15 μl of hybridization solution containing l. 5×104 cpm of the 3H labeled 2.2 kb Ha ras DNA probe was applied to each chromosome spread for overnight hybridization. Banded chromosomes (Wright stain) were compared to standard karyotype of the rat 22 and Levan's nomen clature for banded chromosomes was adopted 13. A slight modification of the method was the use of 95% ethanol containing O. 04% HCl for destaining to prevent loss of silver grains.

G. Extraction and hybridization of nucleic acids

The method of extraction of genomic DNA was based on that of Herrman 24 and that of total RNA preparation on the guanidine hydrochloride procedure of Strohman et al 25. Identification of transfected Ha ras was achieved by Southern blot technique 26. Ten or more μg genomic DNA, digested by restriction enzymes, was hybridized with the probe after eleetrophoresis and transferred to Hybond N membranes (Amersham). Expression of RNA was analyzed by Northern blot technique 27. Denaturation and electrophoresis of RNAs were carried out in the presence of formaldehyde. Glyoxal was used in RNA dot blot experiments.

H. Soft agar assay and tumorigenicity test

Soft agar assay was done in 60 mm culture dishes containing 0.5% agar as the lower layer and 0.33% agarose as the top layer in DMEM–F12–10%FCS medium. Cells were seeded at a density of 2×104 per dish and dishes were monitored for 3 wk. Tumorigenicity tests were performed as follows: confluent cells were harvested and resuspended in Hanks' solution at a concentration of 2×107 cells/ml, and 0.1 ml such suspension was inoculated subcutaneously into the back of each of the ld–old Wistar baby rats. Animals were inspected for 5 m.

Results

A. Establishment and characterization of the LW13 hepatic epithelial cell line

The sequential sedimentation and passage of trypsinized cell suspensions from livers of 10 d-old rats gave rise to 3 sets of flasks with different cell populations. While an epithelial cell population developed in the third set of flasks and was further subcultured, the cells from the first and second sets of flasks were discarded due to their dominance of fibroblast–like cells. Subcultured epithelial cells were of an adherent, polygonal type, growing to flat monolayer (Fig 1). Generally the cells had one nucleus with 1–3 round or oval nucleoli. This epithelial -like growth pattern remained unchanged during long term propagation. The developed cell line was designated as LW13.

Fig 1
figure 1

Morphology of LW13 cells at passage 6 with polygonal shaped cells grown just to confluency. Phase contrast. ×240

Full size image

A growth curve of LW13 cells at passage 28, fed with 10%BS complete medium, is shown in Fig 2. The average doubling time was 39 h. For LW13 ceils at passage 12 the number of confluent cells 15 d after passaging was 6.03×104 cells/cm2, revealing the property of density inhibition of the cells.

Fig 2
figure 2

Growth curve of LW13 cells in 10% BS complete medium.

Full size image

Chromosome analysis at passage 15 showed that from 50 metaphases, 90% were diploid while the rest of the cells had chromosome numbers of 42±1 or 40. The dominant diploid karyotype was stable for more than 50 passages.

LW13 cells maintained the hormonal responsiveness of synthesizing TAT which might be regarded as a kind of hepatocyte specific function. Tab 1 shows the statistically significant increment of TAT activity of LW13 cells after dexamethasone treatment.

Tab 1 Activity of Tyrosine Aminotransferase in LW13 cells
Full size table

TEM not only showed the existence of desmosomes which confirmed the epithelial nature of LW13 cells (Fig 3) but also abundant mitochondria, dilated endoplasmic reticulum (Fig 4) and in particular the intercellular structures resembling bile canaliculi with microvilli limited by junctional complexes (Fig 5). These characteristics are similar to those described for adult rat hepatocyte 28. No virus particles were found.

Fig 3
figure 3

Ultrastructure of LW13 cells showing the presence of desmosomes (D) and microvilli(M). × 17600.

Full size image
Fig 4
figure 4

Ultrastructure of cytoplasm of LW13 cells showing the presence of free mitochondria(M) and dilated rough endoplasmic reticulum(ER). × 11500.

Full size image
Fig 5
figure 5

Ultrastructure of LW13 cells showing the presence of intercellular structure resembling bile canaliculi (C). ×6700.

Full size image

Finally, LW13 cells were shown to be immunostained positively for anti keratin antibodies (Fig. 6). Even at passage 50 the cells were still anchorage dependent and nontumorigenic.

Fig 6
figure 6

The presence of keratin filaments in LW 13 celjls . ×600

Full size image

Based on the above characterization it appears obvious that LWl3 is a normal liver epithelial cell line possessing at the same time some structural and functional properties of rat liver parenchymal cells.

B. Transformation of LW13 cells by exogenous Ha ras gene

After transfection, G418 resistant colonies were visible 7 12 d later in selection medium. An average of approximately 20 colonies appeared among 106 transfected cells. Morphologies of colonies of LW13 cells transfected with vector alone and of many colonies transfected with pSV2 neo Ha ras were similar to those of LW13 cells, but some pSV2 neo Ha ras transfected colonies showed completely different morphology. Both kinds of colonies were then picked up with clone ring and propagated into cell lines. 3 pSV2 neo alone transfected (N3–N5), 4 pSV2 neo Ha ras transfected hut morphologically unchanged (H1–H4), and 9 pSV2 neo Ha ras transformed (HA HI) colonies were analyzed. Fig 7 shows a representative transformed cell line HD. The transformed cells grew in multilayer, proliferated rapidly, and formed foci in soft agar (Fig 8). Typically, cell density of clone HD could reach up to 4.4×105/cm2 which was 7.3 fold as high as normal untransformed LW13 cells in complete medium. These transformed cells could give rise to tumors when inoculated into 1 d old rats, while LW13 cells, pSV2 neo transfected LW13 cells as well as many pSV2 neo Ha ras transfected but morphologically unchanged LW13 cells all failed to produce tumors up to 5 months in vivo. Most of the tumorigenic cells were also capable of producing metastases as shown in Tab 2.

Fig 7
figure 7

Morphology of the pSV2 neo Ha ras transformed malignant cell line (HD). ×240

Full size image
Fig 8
figure 8

A representative focus produced in soft agar by the tumorigenic cell line HA. ×115

Full size image
Tab 2 Summary of transfection experiments
Full size table

Tab 2 summarizes the results obtained from the transfection experiments.

Southern blot analysis confirmed that only pSV2 neo Ha ras transfected cell lines had integrated exogenous ras gene, and that individual clones might have different amounts of copies of the gene (Fig 9). RNA dot blot analysis indicated that only tumorigenic cell lines expressed high levels of Ha ras gene (Fig 10), suggesting that overexpressed Ha -ras was a causal factor of transformation.

Fig 9
figure 9

Detection of exogenous EJ–Ha–ras gene among transfetted cell lines by Southern blot. 15 μg genomic DNA extracted from each cell line was analyzed as described in procedure. laneA, LW13; laneB, N3; laneC, HA; lanesD –J,HC–HI seven cell lines; lanesK–N,H1–H4 four cell lines. Arrowheads on the left site indicate λ–Hind III DNA marker, from top to bottom: 2.0, 2.3, 4.4, 6.6, 9.4, 23. 1 kb respectively; two arrowheads on the right side indicate the exogenous Ha–ras. Serial number of all cell lines are listed in Table 2.

Full size image
Fig 10
figure 10

Experssion of the Ha–ras gene as detected by dot blot, Graded amounts of samples were 10, 2.5, 0. 625 μg total RNA. lane A, LW13; lane B,N3; lane C,N5; lane D, H1; lanesE–I, HG, HI, HF, HD and HA respectively.

Full size image

It was interesting to note that the tumors grown from Ha -ras gene transformed cells (e. g. line HD) contained areas of poorly differentiated hepatocellular carcinomas which consisted of nonuniform cells with nuclei of different sizes and darkly stained nuclear membrane (Fig 11). This indicated that LW13 cells, as a liver epithelial cell line, could contribute to the formation of hepatocellular carcinomas, when transformed by the activated oncogene, Undifferentiated anaplastic carcinoma cells, however, were also found in tumors derived from the same line (Fig 12).

Fig 11
figure 11

Poorly-differentiated hepatocellular carcinoma produced by HD cell line. Hematoxylin stain. ×200

Full size image
Fig 12
figure 12

Anaplastic carcinoma existed in some parts of tumors produced by HD cell line. Hematoxylin stain. ×150.

Full size image

C. Incorporation of exogenous Ha- ras gene

To investigate the relationship between the position of the integrated Ha -ras gene and transformation, a radiolabeled 2.2kb probe which could not be hybridized with rat cellular sequences (Fig 13) was used to localize the human EJ-Ha -ras gene on rat chromosomes by in situ hybridization technique.

Fig 13
figure 13

Identification of homology between the 2.2kb probe, rat cellular Ha -ras and exogenous EJ - Ha - ras. 15μg BamHI digested genomic DNA of each cell line was hybridized with the 2.2kb probe, lane A, LW13; lane B, N3; lane C, N5; lanes D–J, HC, HD,......HI; arrow- heads on the right side indicate λ –Hind III DNA marker, from top to bottom: 2.0, 2.3, 4.4, 6.6, 9.4, 23.1 kb respectively; the left arrowhead indicates 6. 3 kb band.

Full size image

The incorporated sites of EJ -Ha -ras were determined by chromosome banding through emulsion. Karyotype analysis showed that the cell lines HD, HE, HF and HI investigated were aneuploid. But except for the abnormal chromosome No. 13 found in HD cell line and chromosome No. 2 found in HA cell line, there was no other apparent chromosome translocation or band deletion.

In the present work,at least 50 metaphases of each cell line were examined microscopically and the positions of hybridized spots were recorded. Furthermore, according to the altered karyotype of tumorigenic cells, chromosome photographs were pooled and classified into 8 categories, based on their main characteristics (chromosome length and arm ratio). In Tab 3, the chromosomes of 25 metaphases from each of 4 transformed cell lines, HD, HE, HF and HI, were photographed and grouped into 8 categories. For each category, the average length of chromosomes, the number of chromosomes actually counted (in 25 metaphases)etc,were listed. Above all,the percentages of total chromosome length of each category in the total chromosome length of all categories (Tab 3,5) were calculated. In Tab 4, the total number of silver grains found in each 25 in situ hybridized metaphases of 4 different cell lines and the observed number of silver grains in different categories were given. For comparison, the expected number of silver graihs which would represent a random distribution of silver grains were also listed in Tab, 4. A χ2 test was adopted to check up whether the silver grains actually observed in photographs are signals (statistically significant) or only a general background (statistically insignificant comparing to the expected values). The results demonstrated that among the different cell lines HD, HE, HF and HI, most of the hybridized spots accumulate in different categories. However, within one cell line, most of the hybridized spots are confined to one definite category. The χ2 test indicated that an abundance of spots in the definite category reflects a significant signal rather than a general badkground. Apparently, these results demonstrated that for an individual cell line the integration sites are on the same chromosome and this can be explained by its clonal origin.When different cell lines are compared, however, EJ–Ha–ras is found to be incorporated into rat chromosomes in a random pattern.

Tab 3 Classification and data analysis of chromosomes from in situ hybridized metaphases
Full size table
Tab 4 Analysis of the distribution of hybridized silver grains in chromosomes of individual category in 4 transformed cell lines
Full size table

Further investigation into the relatively well banded chromosome spreads has testified the incorporation of the EJ -Ha -ras gene in HD at position lq22. in HE at 4q24,in HF at 7q13. and in HI at 15q25. Fig. 14 and 15 are representatives for hybridized metaphases of HD and HI, respectively. Fig. 16 shows the hybridized spots matched to special region of chromosomes in HD, HF and HI respectively.

Fig 14
figure 14

A typical hybridized metaphase of HD cell, arrowhead indicates the hybridized spots on chromosome No. 1.

Full size image
Fig 15
figure 15

A typical hybridized metaphase of a HI cell; arrowhead indicates the hybridized spots on chromosome No. 15.

Full size image
Fig 16
figure 16

Hybridized chromosomes of cell lines HD, HE and HI.

Full size image

D. Overexpression of endogenous c myc genes coupled with ras-induced transformation

The demonstration of the expression of c-myc gene with a radiolabeled 420bp mouse c myc exon 2 fragment by Northern blot analysis showed an overexpression Of this gene in ras transformed LW13 cells (Fig 17). Densitometer scanning indicated an approximately 10 fold increase of RNA expression of the c myc gene in the tumorigenic cell lines HC, HD, HE, but not in LW 13. pSV2 neo transfected alone, and pSV2 neo Ha ras transfected nontumorigenic cells (Fig 18). Data from Northern blot and dot blot experiments with a radiolabeled 4.8kb c-myc probe (a fragment generated from pSVc myc 15) were consistent with the results presented here (photographs not shown).

Fig 17
figure 17

Overexpression of c myc among Ha ras transformed cell lines. 15μg total RNA from each cell line was analyzed by Northern blot with a 420bp probe (see methods section), laneA, LW13; laneB, N3; laneC, HC; laneD,HD; laneE,HE; arrowhead on the right side indicates 2.5kb c myc band.

Full size image
Fig 18
figure 18

Expression of c myc gene. Signal intensity of Northern blot x ray film was measured by densitometric method with Shimadza dual wavelength thin-layer chromoto-scanner model CS 930. 15μg total RNA for each cell line was analyzed by Northern blot with 420bp probe, column1, LW13; column 2, N3; column 3, HC; column 4, HD; column 5, HE; column 6, H1.

Full size image

Discussion

This paper describes a series of experiments, including the establishment of a rat liver epithelial cell line LW13, in vitro transformatin of LW13 cells by the EJ Ha ras gene,the detection of integrated exogenous ras gene and of expression of endogenous c myc gene. The results obtained provided additional information concerning in vitro transformation and hepatocarcinogenesis in the rat.

First of all, it seems essential to confirm the hepatic origin of LW13 cell line. The sequential sedimentation method used here permitted the attachment of hepatocytes as well as the nonparenchymal epithelial cells, but the fibroblasts were eliminated. The LW13 cells maintained the inducible potential to express hepatic functions, e.g. the production of TAT. Concerning the ultrastructures, the features of rough endoplasmic reticulum (RER), abundance of mitochondria, and particularly intercellular structures resembling bile canaliculi were similar to the characteristics of hepatocytes 28. In this paper, however, LW13 cells were regarded rather as nonparenchymal cells,taking the following aspects into account: 1) neither the extent of the RER nor the amount of mitochondria of LW13 cells is as high as in hepatocytes; 2) it is generally agreed that nonparenchymal epithelial cells are easier to develop into cultures whereas hepatocytes can survive not longer than several weeks in culture; 3) they are smaller than parenchymal cells . We, therefore, considered LW13 cells as rat liver nonparenchymal epithelial cells having some characteristics of parenchymal cells.

Until recently it had widely been believed that hepatocytes are the origin of hepatoma cells. However, Fausto et al 29, 30 pointed out that a kind of nonparenchymal epithelial cells, so called “oval cells”, could be transformed, yielding moderately to well differentiated hepatocellular carcinomas after transfection by EJ Ha ras. Interestingly, these tumor cells could even express antigens of both hepatocytes as well as oval cells 31. Tsao et al 32 described a diploid epithelial cell line from rat liver bearing phenotypic properties of oval cells. Although the exact relationship between hepatocytes, nonparenchymal oval cells and culturable epithelial cells such as LW13 remains to be further elucidated, the fact that LWl3 cells can be transformed in vitro, resulting in cells capable of producing tumors with areas of poorly differentiated hepatocellular carcinomas in vivo would support the hypothesis that nonparenchymal epithelial cells may contribute to the genesis of hepatocellular carcinomas in vivo under certain conditions.

With respect to the specific integration of exogenous ras gene in rat chromosomes as claimed by some authors 12, our data of in situ hybridization demonstrated only a random incorporation of exogenous EJ Ha ras. The existence of specific sequences which may have the potential to facilitate the integration of transfected ras gene can not be excluded, although we have never found the incorporation of a human ras gene into rat chromosome No. 3.

One of the consequences of the in vitro transformation of LW13 cells is the overexpression of endogenous c myc gene. It is quite possible that c myc plays an important role during the ras induced oncogenesis. To investigate the functional relationship between ras and myc genes, c myc antisense RNA work has been carried out in further experiments by the present authors (unpublished data). A fragment of the mouse c myc gene was ligated into pSV2 plasmid in opposite orientation and was introduced into the ras transformed LW13 cells. Southern blot experiments confirmed the integration of this antisense c myc DNA sequences. Delayed growth initiation in low serum medium of antisense RNA expressing cells has been observed, indicating the influence of c myc gene expression on the malignant phenotype of LW13 cells transformed by the ras gene. A modification of the present system has been developed for further investigation of the c myc gene function.